A PR masterclass from Anthropic’s Claude Mythos Preview: power, control and risk

This is the latest in a series of articles to celebrate PRmoment's PR Masterclass: AI in PR, where we talk to experts about how AI is changing the PR sector. If you want to understand more about the how AI is changing the output and make up of PR teams, often for the better, come along to PR Masterclass: AI in PR. Everyone love's it!

In high fashion, scarcity creates desire. In frontier AI, it creates something more complicated: desire, suspicion, and regulatory attention. Anthropic’s Claude Mythos Preview has not been widely released, yet governments, regulators, and financial institutions are already treating it as consequential. Restricted access has not limited scrutiny – it has helped create it.

It’s already being framed as a PR strategy – that reading is useful, but the implications go further. When a product is judged before it is broadly available, communications does more than explain the technology. It helps set the terms on which audiences assess it: whether they see a breakthrough, a risk, or a company that is in control.

The Anthropic edge

In a market where most audiences cannot independently verify performance, perception becomes an early competitive advantage. If buyers, investors, and policymakers cannot test the product for themselves, they look for proxies: who gets access, who is concerned, and how seriously the company appears to be managing risks.

Anthropic appears to have understood that dynamic. By presenting Mythos as powerful but tightly controlled, Anthropic casts itself as both frontier innovator and a responsible actor. That carries weight, particularly with enterprise buyers, financial organisations, regulators, and investors trying to distinguish between ambition and recklessness.

The regulatory response shows why. The EU is reportedly in talks with Anthropic over Mythos, amid concern about its cybersecurity implications, including what advanced vulnerability discovery could mean for banks and critical infrastructure. Mythos is therefore a test case in how communications can create strategic advantage while increasing regulatory exposure.

Limited access can make a product feel more important. In AI, it can also suggest responsibility: the company is not simply throwing a powerful system into the world and hoping for the best.

But scarcity creates its own risks. If a company builds its story around control, any challenge to that control becomes reputationally significant. Reports of unauthorised access to Mythos, investigated by Anthropic, show the tension. A contained technical or operational issue becomes a test of corporate credibility when the public narrative rests on managed access.

For communications leaders, the implication is that trust is becoming operational. It is not a message or a tone of voice. It depends on sustained alignment between narrative and reality, especially where safety and control are central to the story.

For PR and marketing professionals, five lessons follow:

1. Scarcity signals capability, but it also commits a company to control: If access is positioned as tightly managed, the systems behind that access need to withstand scrutiny. Any gap becomes a credibility issue.
2. Safety-led positioning raises the burden of proof: It can be powerful, but it lowers tolerance for failure. The more responsibly a company claims to behave, the faster it needs to respond when something goes wrong.
3. Narrative can create regulatory momentum: The way a company frames risk can help establish the importance of a product, but it can also accelerate external intervention. Once governments start asking questions, market positioning becomes a public-interest issue.
4. Visibility needs validation: If the story is running ahead of broad access, companies need an evidence plan in place before scrutiny peaks: third-party testing, customer pilots, independent benchmarks and clear access criteria. The point is not to support the launch story – it is to substantiate it under pressure.
5. Language becomes the yardstick: The standards a company sets publicly will become the standards used to assess it. Claims about control and safety are not just positioning – they are commitments. They need to be chosen carefully, defined clearly and understood internally.

The lesson extends beyond Anthropic, and beyond AI. As technology becomes harder for non-specialists to evaluate, communications increasingly shapes which companies are believed early, and which are challenged first. It determines who gets the benefit of the doubt.

The opportunity is early influence, and the trade-off is earlier accountability. In frontier technology, the story is not just something added at launch. It helps decide who is trusted before the evidence is widely available – and who is held to account when the evidence arrives.

If this article has inspired you to learn more about how AI will impact the PR agency model, don't miss PRmoment's PR Masterclass: AI in PR. Our no bluffers policy, no sponsors policy: The speaker line-up is packed with insight from people who are actually doing it. Not those who just write (endless?) LinkedIn posts about it. And we don't have any sponsors on the speaker programme; our editorial team choose the speakers, not our commercial team!